APPLYING DATA ENVELOPMENT ANALYSIS AND CLUSTERING ANALYSIS IN ENHANCING THE PERFORMANCE OF PHILIPPINE NATIONAL POLICE-DISTRICT VI IN THE PROVINCE OF CAVITE
The document summarizes a study that used data envelopment analysis and clustering analysis to evaluate the performance of police stations in Cavite, Philippines and identify crime patterns. Data envelopment analysis was used to measure the efficiency of four police stations based on inputs like staffing and outputs like crime solved. K-means clustering analyzed crime data to group similar crimes and identify trends over 2014-2016. The analysis found theft to be the most common crime from 2014-2015 while murder was most common in 2016. The tools helped identify how resources could be optimized and predict future crime trends.
IRJET- Algorithms for the Prediction of Traffic Accidents
Similar to APPLYING DATA ENVELOPMENT ANALYSIS AND CLUSTERING ANALYSIS IN ENHANCING THE PERFORMANCE OF PHILIPPINE NATIONAL POLICE-DISTRICT VI IN THE PROVINCE OF CAVITE
Correlation Method for Public Security Information in Big Data EnvironmentIJERA Editor
Similar to APPLYING DATA ENVELOPMENT ANALYSIS AND CLUSTERING ANALYSIS IN ENHANCING THE PERFORMANCE OF PHILIPPINE NATIONAL POLICE-DISTRICT VI IN THE PROVINCE OF CAVITE (20)
Software and Systems Engineering Standards: Verification and Validation of Sy...
APPLYING DATA ENVELOPMENT ANALYSIS AND CLUSTERING ANALYSIS IN ENHANCING THE PERFORMANCE OF PHILIPPINE NATIONAL POLICE-DISTRICT VI IN THE PROVINCE OF CAVITE
1. International Journal of Information Technology, Control and Automation (IJITCA) Vol. 7, No.2, April 2017
DOI:10.5121/ijitca.2017.7201 1
APPLYING DATA ENVELOPMENT ANALYSIS
AND CLUSTERING ANALYSIS
IN ENHANCING THE PERFORMANCE OF
PHILIPPINE NATIONAL POLICE-DISTRICT VI
IN THE PROVINCE OF CAVITE
Mengvi P. Gatpandan , Shaneth C. Ambat
School of Graduate Studies, AMA University,Quezon City, Philippines
ABSTRACT
Data envelopment analysis is a technique or method for assessing and evaluating the relative performance
of organizational entities where the manifestation of multiple inputs and outputs makes comparison
difficult. Efficiency was measured through data envelopment analysis in Philippine National Police District
VI in the Province of Cavite to measure the performance of decision-making units in terms of their
resources. Clustering is the process of grouping and analyzing the list of objects which have similar
characteristics. Clustering algorithm is used in this study to help identify crime pattern. The clustering
algorithm was implemented in the application software Crime Management System (CriMS) to predict the
crime pattern to help Philippine National Police District VI in the Province of Cavite in decreasing the
total number of crime volume and increase the number of crimes solved to countervail security concerns of
an individual, community, and the state. Further studies must be conducted to determine the usefulness of
the application software by leading to an empirical study on the rule set used to determine the predictive
accuracy and/or software productivity.
KEYWORDS
Data Envelopment Analysis, Data Mining, Clustering Techniques, K-means Algorithm,
1. INTRODUCTION
The population shift from rural to urban areas has focused the lenses of urbanization such as the
endogenous, modernization and world system. This shift has tremendously posed enormous
challenges and pressing serious problems which considerably threaten security issues due to
rising urban crimes, expansion of slums, environmental degradation: pollution and vulnerability
to flooding [1] [2].
In this changing society, the continuous measures to resist unlawful acts punishable by a state
which in any form an act harmful not only to a certain individual but also to the society or to the
state. The Philippine Development Plan 2011-2016 recognizes the need for a safe and secure
environment as an important factor in fostering investment and the country's economic growth.
The protection of the state public order and stability envisions casting out the index and high
profile crimes are consistent with the R.A. No. 3815 or the Revised Penal Code of the Philippines
which penalizes crimes committed against legal code or law.
The government has the sole power by the mandate of its legislations and crime regulatory bodies
to severely restrict one's liberty for the commission of crimes. Philippine National Police (PNP)
has defined crime classification as an index and non-index crimes. Index crimes involve 1) crimes
against persons such as a) murder, b) homicide, c) physical injury and d) rape, and 2) crimes
against property such as a) robbery, b) theft, c) car-napping/carjacking and, d) cattle rustling
while non-index crimes are violations of special and private laws such as local ordinances.
2. International Journal of Information Technology, Control and Automation (IJITCA) Vol. 7, No.2, April 2017
2
Data envelopment analysis (DEA) is a technique or method for assessing and evaluating the
relative performance of organizational entities where the manifestation of multiple inputs and
outputs make comparison difficult [3]. Norman & Stoker [4] noted that DEA is a powerful,
analytical technique for evaluating the performance of organizational units in the private or public
sector. On their critics and review, Lertworasirikul, Fang, Joines, & Nuttle [5] posited that “data
envelopment analysis (DEA) models require crisp input/output data especially in evaluating the
performance of activities or organizations services”. Charnes, Cooper, & Rhodes [6] used DEA in
measuring the efficiency of decision-making units.
Berry & Linoff [7] defines data mining acquaints new methodologies and techniques for a better
and informed decision through a careful examination of large databases and considered as
knowledge mining from data [8] towards the development of models about aggregated data [9].
A cluster is a group of objects with similar attributes or object characteristics. Data Clustering is
the process of grouping and analyzing the list of objects which have similar characteristics. It
was exemplified [10] that clustering is "unsupervised learning" and a technique for grouping
similar data points [11]. Recent scholarly works show that data mining can aid in crime
detection problems and speed up the crime resolution. A clustering algorithm helps identify crime
patterns and helps to improve in decreasing the number of crime to countervail security concerns
of an individual, community, and the state.
The researchers expressed and presented new detection through the criminological enterprise of
crime rampant to the Philippine National Police – District VI in the Province of Cavite to aid the
organization and the community. The researchers asked permission to conduct the research study
in the Cavite Provincial Police Office (Cavite PPO) to the Officer-in-charge through the Chief of
PIDMB that handled the Statistics on Criminal Cases of Index Crime of all Police Station in the
Province of Cavite. The historical data available in Cavite PPO and lenses of the research study
started from the year 2014 up to the year 2016.
2. LITERATURE REVIEW
Several scholarly works about measuring the efficiency using data envelopment analysis in
different organization of DMUs were used. Chan & Karim [12] used a two-stage estimation DEA
technique in East Asian economies to determine the relationship between the financial market
regulation, country governance, and efficiency of commercial banks during the period of 2001-
2008. Kinachi, Najjari, & Alp [13] used DEA and stochastic frontier analysis methods for the
scores efficiency and hydroelectricity centers rank to measure the efficiency of 32 Iranian
electricity industry. The study used an input-output oriented model using CCR-model and BCC-
model in DEA. Osman, Berbary, Sidani, Al-Ayoubi, & Emrouznejad [14] focused on the
performance and appraisal evaluation for nurses using a data envelopment analysis. The
assessment and relative performance of nurses was and useful for both nurses and hospital in the
age of clinical supremacy. Ulucan [15] measured the efficiency in higher education institutions
using DEA in Turkish universities using multiple inputs and outputs. Estrada, Song, Kim, Namn,
& Kang [16] focused on a dynamic method in benchmarking to identify and measure the
inefficient DMUs to improve the efficiency progressively from the dataset collected from the
Canadian Bank branches. The study proposed an active method of stepwise benchmarking for
the inefficient DMUs.
Predictive analysis uses data mining techniques. Data mining is extracting hidden knowledge,
useful and meaningful pattern and trends in a large data set in which organization uses it for
decision-making purpose. Grubesic, Wei, & Murray [17] Cluster analysis continues to be an
important exploratory technique in scientific inquiry. Several scholarly works used it widely in
public health, ecology, geography, and many other fields. Clustering data mining technique can
be used for crime detection and crime prevention. Jin-ho & Seung-Ryul [18] used text mining to
extract and treat useful information based on the natural language processing, opinion mining to
3. International Journal of Information Technology, Control and Automation (IJITCA) Vol. 7, No.2, April 2017
assigned positive and negative or neutral preference to social media unstructured data, and social
network analysis to discover viral objects by measuring a user’s reputation or influence based on
their connection network and clustering analysis. Ceccato, & Uittenbogaard [19] assessed the
crime rates in underground stations using space
crime. Tayebi, Ester, Glasser, & Brantingham [20] specialized on s
focused on crime hotspot areas with disproportionality in which location has higher crime
density. Rajagopal [21] research study is all about customer clustering. Segmentation is one of
the most important factors used in the st
model for density change among spatial regions using density tracing
large aggregated crime datasets.
identify the gait pattern classification.
3. METHODOLOGY
The research study employs the quantitative method of research. The quantitative research
involves formal, objective information with mathematical quantification. During the
developmental phase, Cross Industry
model will be utilized.
3.1. RESEARCH DESIGN
Data Envelopment Analysis was used to determine and assess the performance efficiency of the
four (4) Police Stations of District VI in the province of
organizational resources such as manpower, physical, financial and technology, and the total
volume of index crimes committed per year.
using the illustrations below:
Data Clustering is the process of grouping and analyzing the list of objects which have similar
characteristics. The research study will use the partition as the clustering technique and the k
means method as the clustering algorithm. Devi &
to cluster observations into groups or clusters of associated observations without any prior idea or
knowledge of those relationships. According to
algorithm works as follows:
1. Initialize the center of the cluster
2. Attribute the closest cluster to each data point
3. Set the position of each to the mean of all data points belonging to that cluster
4. Repeat Steps 2-3 until convergence.
International Journal of Information Technology, Control and Automation (IJITCA) Vol. 7, No.2, April 2017
assigned positive and negative or neutral preference to social media unstructured data, and social
network analysis to discover viral objects by measuring a user’s reputation or influence based on
ir connection network and clustering analysis. Ceccato, & Uittenbogaard [19] assessed the
crime rates in underground stations using space-time variation analysis to extract patterns for the
crime. Tayebi, Ester, Glasser, & Brantingham [20] specialized on spatial crime analysis that
focused on crime hotspot areas with disproportionality in which location has higher crime
density. Rajagopal [21] research study is all about customer clustering. Segmentation is one of
the most important factors used in the study of marketing. Phillips & Lee [22] used to develop a
model for density change among spatial regions using density tracing-based approach for the
large aggregated crime datasets. Semwal, Vijay Bhaskar, et al., [23] used clustering analysis to
he gait pattern classification.
The research study employs the quantitative method of research. The quantitative research
involves formal, objective information with mathematical quantification. During the
developmental phase, Cross Industry Standard Process for Data Mining (CRISP-DM) process and
Data Envelopment Analysis was used to determine and assess the performance efficiency of the
four (4) Police Stations of District VI in the province of Cavite as DMUs in terms of
organizational resources such as manpower, physical, financial and technology, and the total
volume of index crimes committed per year. DEA assess and evaluate the efficiency of DMUs
Eq. (1)
Data Clustering is the process of grouping and analyzing the list of objects which have similar
characteristics. The research study will use the partition as the clustering technique and the k
means method as the clustering algorithm. Devi & Rajagopalan [24] k-means clustering is used
to cluster observations into groups or clusters of associated observations without any prior idea or
knowledge of those relationships. According to Seddawy, Khedr, & Sultan [25]
Initialize the center of the cluster
Eq. (2)
Attribute the closest cluster to each data point
Eq. (3)
Set the position of each to the mean of all data points belonging to that cluster
Eq. (4)
3 until convergence.
International Journal of Information Technology, Control and Automation (IJITCA) Vol. 7, No.2, April 2017
3
assigned positive and negative or neutral preference to social media unstructured data, and social
network analysis to discover viral objects by measuring a user’s reputation or influence based on
ir connection network and clustering analysis. Ceccato, & Uittenbogaard [19] assessed the
time variation analysis to extract patterns for the
patial crime analysis that
focused on crime hotspot areas with disproportionality in which location has higher crime
density. Rajagopal [21] research study is all about customer clustering. Segmentation is one of
udy of marketing. Phillips & Lee [22] used to develop a
based approach for the
Semwal, Vijay Bhaskar, et al., [23] used clustering analysis to
The research study employs the quantitative method of research. The quantitative research
involves formal, objective information with mathematical quantification. During the
DM) process and
Data Envelopment Analysis was used to determine and assess the performance efficiency of the
Cavite as DMUs in terms of
organizational resources such as manpower, physical, financial and technology, and the total
DEA assess and evaluate the efficiency of DMUs
Eq. (1)
Data Clustering is the process of grouping and analyzing the list of objects which have similar
characteristics. The research study will use the partition as the clustering technique and the k-
means clustering is used
to cluster observations into groups or clusters of associated observations without any prior idea or
Seddawy, Khedr, & Sultan [25], k-means
Set the position of each to the mean of all data points belonging to that cluster
4. International Journal of Information Technology, Control and Automation (IJITCA) Vol. 7, No.2, April 2017
4
Notation: lcl = number of elements in c
A CRISP-DM methodology was used as a research design of the study. CRISP-DM provides a
structured approach to planning a data mining project in the analytical task in the process of
criminal analysis. The researchers used the data mining framework shown below.
Figure 1.0 CRISP-DM Model
Business Understanding. It focuses on understanding and identifying the project objectives and
requirements from a business perspective, and converts the knowledge in a data mining problem
definition and provides initial plan design to achieve the desired objectives. The researchers’
goal is to help the Philippine National Police – District VI in the Province of Cavite to decrease
the index crime rate.
Data Understanding. Initial data gathered and collected from the four (4) police stations at Cavite
Provincial Police Office (PPO) situated at Imus, Cavite then determines data quality problem to
discover interesting subsets to form an assumption for hidden information.
Data Preparation. Identify, select and prepare data attributes needed to data mining and cleanse
data for modeling tools.
Modeling. Selection of clustering modeling techniques, generate test design, form, and assess the
model. The k-means algorithm was used for identifying the index crime patterns.
Evaluation. Evaluate and check the model results if it generates the desired results of the study.
Deployment. Provide an organized and presented data report for the intended beneficiaries which
includes detailed findings, explanation of models, and others to discuss the initial data mining
goals have been met.
3.2 OTHER TOOLS USED IN THE RESEARCH STUDY
MaxDEA software was used to obtain measures of productivity and efficiency to conduct data
envelopment analysis. Also, SQL server 2012 with Microsoft Excel 2013 was used to determine
crime pattern using clustering technique. Further, Microsoft.Net 4.5 was used for environment
framework, Visual Studio 2013 for integrated development environment, Visual C# as
programming language, ASP.Net MV5 as development framework for single page application in
the web-based environment, SQL server 2012 for data store/persistence, Internet Information
Services for web server, jquery 1.10 for javascript framework, jquery easyui for front-end user
interface framework, google charts and OLAP pivot graph for charting and visualization, Entity
framework 6 for domain entities/models and dapper micro-form for access.
4. RESULTS AND DISCUSSION
The efficiency of the Philippine National Police – District VI was determined through Data
Envelopment Analysis using Radial Measure of Efficiency with Input-oriented orientation and a
5. International Journal of Information Technology, Control and Automation (IJITCA) Vol. 7, No.2, April 2017
5
Technical Scale Technical Scale Technical Scale
Efficiency Efficiency Efficiency Efficiency Efficiency Efficiency
Score Score Score Score Score Score
Municipality 1 0.27 0.27 0.59 0.59 1.00 1.00 0.62 3
Municipality 2 0.82 0.82 1.00 1.00 0.72 0.72 0.85 2
Municipality 3 1.00 1.00 1.00 1.00 1.00 1.00 1.00 1
Municipality 4 1.00 1.00 0.56 0.67 0.94 0.94 0.85 2
RankingDMU
2014 2015 2016
Average
Values Probability Values Probability Values Probability
CrimeCategory THEFT 34% MURDER 30% RAPE 24%
CrimeCategory ROBBERY 28% THEFT 17% PHYSICALINJURIES 24%
CrimeCategory MURDER 16% RAPE 17% MURDER 18%
CrimeCategory PHYSICALINJURIES 9% ROBBERY 17% THEFT 18%
CrimeCategory RAPE 6% PHYSICALINJURIES 13% MOTORNAPPING 12%
CrimeCategory CARNAPPING 3% MOTORNAPPING 4% ROBBERY 6%
CrimeCategory HOMICIDE 3%
2016
Population (All)
Variables
2014 2015
Rate-to-Scale (RTS) using Scale efficiency. The same result was obtained using input-oriented
and output-oriented Chames, Cooper and Rhodes (CCR) model.
Table 1.0 : The variables used in measuring the efficiency of the PNP
Table 1.0 presents the variables used in measuring the efficiency of the Philippine National Police
District VI such as Municipality 1, Municipality 2, Municipality 3 and Municipality 4 as DMUs,
personnel, mobile units, radio, computer, printer, budget, firearms and total crimes as inputs, and
a number of crimes cleared and crimes solved as output.
Table 2.0 : 3-Year Scale Efficiency Results of the PNP – District VI in the province of Cavite
Table 2.0 presents the 3-year Scale Efficiency Results of the Philippine National Police – District
VI in the province of Cavite. Municipality 1 got the average score of 0.62 scale efficiencies and
ranked as the lowest inefficient DMU; Municipality 2 and Municipality 4 got the average score of
0.85 scale of efficiency and ranked as 2nd
inefficient DMU. Among the 4 DMUs, Municipality 3
got an average score of 1.00 scale of efficiency in 3 years. Therefore, Municipality 3 was the
efficient DMU of the Philippine National Police – District VI in the Province of Cavite.
The following shows the clustering technique using data mining tools in MS SQL Server 2012
using Microsoft Excel 2013 with the raw data of criminal cases of District VI for three (3) years
historical data from January 2014 up to December 2016.
Table 3.0 : Cluster Characteristics of Municipality 1 by Crime Category Year 2014 – 2016
Table 3.0 shows the cluster characteristics of Municipality 1 by Crime Category year 2014 –
2016. The summary shows that theft ranked the highest in the year 2014 with 34% probability,
murder in the year 2015 with 30% probability and rape in the year 2016 with 24% probability.
Rule Set for Year 2016
Cluster 1: CrimeCategory=PHYSICALINJURIES, CrimeCategory=RAPE,
6. International Journal of Information Technology, Control and Automation (IJITCA) Vol. 7, No.2, April 2017
6
Values Probability Values Probability Values Probability
Barangay Buho 9% Talon 13% Talon 24%
Barangay Maitim I 9% Barangay II 9% Halang 12%
Barangay Pangil 9% Maitim I 9% Buho 12%
Barangay Maymangga 6% Halang 9% Bucal 12%
Barangay Loma 6% Buho 9% Barangay III 6%
Barangay Halang 6% Salaban 9% Dagatan 6%
Barangay Barangay VII 6% Barangay XI 9% Barangay XI 6%
Barangay Barangay VI 3% Pangil 4% Tamacan 6%
Barangay Talon 3% Barangay V 4% Salaban 6%
Barangay Poblacion 6 3% Loma 4% Barangay IV 6%
Barangay Barangay VIII 3% Barangay X 4% Cabuco 6%
Barangay Salaban 3% Tamacan 4%
Barangay Dagatan 3% Barangay VII 4%
Barangay Barangay IX 3% Dagatan 4%
Barangay Poblacion 5 3% Barangay III 4%
Barangay Poblacion 10 3%
Barangay Banaybanay 3%
Barangay Tamacan 3%
Barangay Barangay I 3%
Barangay Barangay XII 3%
Barangay Barangay II 3%
Barangay Minantok Kanluran 3%
2016
Population (All)
Variables
2014 2015
Values Probability Values Probability Values Probability
MonthComttd July 13% April 17% August 18%
MonthComttd November 13% August 17% April 18%
MonthComttd March 9% November 13% September 12%
MonthComttd December 9% December 9% June 12%
MonthComttd January 9% June 9% May 12%
MonthComttd April 9% September 9% January 12%
MonthComttd August 6% February 9% March 12%
MonthComttd September 6% March 4% November 6%
MonthComttd October 6% October 4%
MonthComttd May 6% July 4%
MonthComttd February 6% January 4%
MonthComttd June 6%
2016
Population (All)
Variables
2014 2015
CrimeCategory=MOTORNAPPING
Cluster 2: CrimeCategory=MURDER, CrimeCategory=THEFT,
CrimeCategory=ROBBERY
Table 4.0 : Cluster Characteristics of Municipality 1 by Barangay Year 2014 – 2016
Table 4.0 shows the cluster characteristics of Municipality 1 by Barangay year 2014 – 2016. The
summary shows that Buho ranked the highest in the year 2014 with 9% probability and Talon in
the year 2015 and 2016 with 13% and 24% probability.
Table 5.0 : Cluster Characteristics of Municipality 1 by Seasonal (Month) Year 2014 – 2016
Table 5.0 shows the cluster characteristics of Municipality 1 by Seasonal (Month) 2014 – 2016.
The summary shows that July ranked the highest in the year 2014 with 13% probability, April and
August in the year 2015 and 2016 with 17% and 18% probability.
Rule Set for Year 2014
Cluster 5: MonthComttd=June, MonthComttd=August, MonthComttd=February,
MonthComttd=January, MonthComttd=October, MonthComttd=September
Cluster 1: MonthComttd=July, MonthComttd=March
Cluster 4: MonthComttd=December, MonthComttd=January, MonthComttd=August,
MonthComttd=September, MonthComttd=June, MonthComttd=May
Rule Set Year 2015
Cluster 4: MonthComttd=November, MonthComttd=February, MonthComttd=June,
7. International Journal of Information Technology, Control and Automation (IJITCA) Vol. 7, No.2, April 2017
7
Values Probability Values Probability Values Probability
CrimeCategory THEFT 39% THEFT 28% THEFT 31%
CrimeCategory PHYSICALINJURIES 24% PHYSICALINJURIES 22% ROBBERY 20%
CrimeCategory ROBBERY 15% ROBBERY 20% PHYSICALINJURIES 18%
CrimeCategory MURDER 10% MURDER 10% MURDER 10%
CrimeCategory RAPE 9% RAPE 8% MOTORNAPPING 8%
CrimeCategory CARNAPPING 2% MOTORNAPPING 5% RAPE 6%
CrimeCategory HOMICIDE 1% CARNAPPING 4% CARNAPPING 5%
CrimeCategory HOMICIDE 2% HOMICIDE 2%
2016
Population (All)
Variables
2014 2015
MonthComttd=March, MonthComttd=January
Cluster 1: MonthComttd=April, MonthComttd=July, MonthComttd=December
Cluster 3: MonthComttd=September, MonthComttd=October, MonthComttd=February,
MonthComttd=July, MonthComttd=January, MonthComttd=June,
MonthComttd=March, MonthComttd=December
Cluster 2: MonthComttd=August, MonthComttd=November
Table 6.0 : Cluster Characteristics of Municipality 2 by Crime Category Year 2014 - 2016
Table 6.0 shows the cluster characteristics of Municipality 2 by Crime Category year 2014 –
2016. The summary shows that Theft ranked the highest in the year 2014 - 2016 with 39%, 28%,
and 31% probability.
Rule Set for Year 2014
Cluster 1: CrimeCategory=THEFT
Cluster 2: CrimeCategory=PHYSICALINJURIES, CrimeCategory=MURDER
Cluster 4: CrimeCategory=HOMICIDE, CrimeCategory=CARNAPPING,
CrimeCategory=RAPE, CrimeCategory=MURDER, CrimeCategory=ROBBERY
Cluster 3: CrimeCategory=ROBBERY, CrimeCategory=RAPE, CrimeCategory=MURDER
Rule Set Year 2015
Cluster 3: CrimeCategory=PHYSICALINJURIES, CrimeCategory=RAPE,
CrimeCategory=CARNAPPING, CrimeCategory=MURDER,
CrimeCategory=HOMICIDE
Cluster 1: CrimeCategory=THEFT, CrimeCategory=MOTORNAPPING
Cluster 2: CrimeCategory=ROBBERY
Rule Set Year 2016
Cluster 3: CrimeCategory=ROBBERY, CrimeCategory=MOTORNAPPING,
CrimeCategory=RAPE, CrimeCategory=CARNAPPING,
CrimeCategory=HOMICIDE
Cluster 1: CrimeCategory=THEFT
Cluster 2: CrimeCategory=PHYSICALINJURIES, CrimeCategory=MURDER,
CrimeCategory=CARNAPPING, CrimeCategory=HOMICIDE,
CrimeCategory=MOTORNAPPING
8. International Journal of Information Technology, Control and Automation (IJITCA) Vol. 7, No.2, April 2017
8
Values Probability Values Probability Values Probability
Barangay Bacao I 9% Manggahan 9% Bacao I 10%
Barangay Manggahan 7% Bacao I 9% Manggahan 10%
Barangay Alingaro 6% Tapia 6% Panungyanan 5%
Barangay Tejero 5% Tejero 6% Tapia 5%
Barangay San Francisco 4% Javalera 4% Javalera 5%
Barangay Tapia 4% Biclatan 4% Alingaro 4%
Barangay Buenavista III 4% San Francisco 4% Tejero 4%
Barangay Arnaldo 4% Prinza 4% Sta. Clara 3%
Barangay Santiago 3% Panungyanan 4% Bacao II 3%
Barangay Bacao II 3% Santiago 3% Vibora 3%
Barangay Pinagtipunan 3% San Gabriel 3% Biclatan 3%
Barangay Bagumbayan 3% Bacao II 3% Navarro 3%
Barangay Navarro 3% Alingaro 3% San Francisco 3%
Barangay Dulong Bayan 3% Governor Ferrer 3% Santiago 3%
Barangay Biclatan 3% San Juan I 3% San Gabriel 3%
Barangay Javalera 3% Buenavista I 3% Prinza 3%
Barangay Prinza 3% Arnaldo 3% Buenavista II 2%
Barangay Buenavista II 3% Buenavista III 2% Arnaldo 2%
Barangay Vibora 3% Pasong Camachile I 2% Governor Ferrer 2%
Barangay Sampalucan 3% Corregidor 2% Corregidor 2%
Barangay Governor Ferrer 2% Pasong Kawayan II 2% San Juan II 2%
Barangay Pasong Camachile I 2% Pasong Camachile II 2% Pasong Camachile II 2%
Barangay San Juan I 2% Sampalucan 2% Pasong Camachile I 2%
Barangay 96th 2% Bagumbayan 2% San Juan I 2%
Barangay Corregidor 2% Navarro 2% 96th 2%
Barangay Panungyanan 2% Pinagtipunan 2% Buenavista I 1%
Barangay San Gabriel 2% Buenavista II 2% Sampalucan 1%
Barangay Pasong Camachile II 1% San Juan II 2% Buenavista III 1%
Barangay San Juan II 1% Vibora 2% Pasong Kawayan I 1%
Barangay Sta. Clara 1% Pasong Kawayan I 1% Dulong Bayan 1%
Barangay Buenavista I 1% Dulong Bayan 1% Bagumbayan 1%
Barangay Pasong Kawayan II 1% 96th 1% Pasong Kawayan II 1%
Barangay Pasong Kawayan I 1% Sta. Clara 1% Pinagtipunan 1%
2016
Population (All)
Variables
2014 2015
Table 7.0 : Cluster Characteristics of Municipality 2 by Barangay Year 2014 – 2016
Table 7.0 shows the cluster characteristics of Municipality 2 by Barangay 2014 – 2016. The
summary shows that Bacao ranked the highest in the year 2014 - 2016 with 9%, 9%, and 10%
probability.
Rule Set for Year 2014
Cluster 4: Barangay=Tapia, Barangay=Pasong Camachile I, Barangay=Prinza,
Barangay=Buenavista II, Barangay=Tejero, Barangay=Javalera,
Barangay=Bacao II, Barangay=Sta. Clara, Barangay=96th, Barangay=San Juan I,
Barangay=Corregidor, Barangay=Bagumbayan, Barangay=Governor Ferrer,
Barangay=Buenavista I, Barangay=Pasong Kawayan II, Barangay=Mangahan,
Barangay=Buenvavista III, Barangay=Vibora, Barangay=Pasong Kawayan,
Barangay=San Gabriel
Cluster 1: Barangay=Pinagtipunan, Barangay=Dulong Bayan, Barangay=Bacao I,
Barangay=Buenavista III, Barangay=Buenavista II, Barangay=Panungyanan,
Barangay=Sampalucan, Barangay=San Juan II, Barangay=Mangahan,
Barangay=Pasong Camachile II, Barangay=Governor Ferrer,
Barangay=Pasong Kawayan II
Cluster 3: Barangay=Arnaldo, Barangay=Biclatan, Barangay=Santiago,
Barangay=San Francisco, Barangay=Vibora, Barangay=Pasong Kawayan I,
Barangay=San Gabriel, Barangay=Pasong Camachile II, Barangay=Panungyanan,
Barangay=San Juan I, Barangay=Pasong Kawayan, Barangay=Sampalucan,
Barangay=Sta. Clara, Barangay=Buenvavista III, Barangay=Tejero
Cluster 2: Barangay=Manggahan, Barangay=Alingaro, Barangay=Navarro,
Barangay=San Juan II, Barangay=Sampalucan,
Barangay=Bagumbayan, Barangay=Tapia
9. International Journal of Information Technology, Control and Automation (IJITCA) Vol. 7, No.2, April 2017
9
Rule Set for Year 2015
Cluster 8: Barangay=Arnaldo, Barangay=Pasong Camachile II, Barangay=Buenavista III,
Barangay=Pinagtipunan, Barangay=Governor Ferrer, Barangay=Bacao II,
Barangay=Panungyanan,Barangay=Pasong Kawayan II, Barangay=Pasong Kawayan
I, Barangay=Navarro, Barangay=San Juan II, Barangay=Corregidor,
Barangay=Sampalucan, Barangay=San Gabriel, Barangay=Bagumbayan,
Barangay=Sta. Clara
Cluster 6: Barangay=Pasong Kawayan II, Barangay=Corregidor, Barangay=Navarro,
Barangay=Buenavista II, Barangay=Vibora, Barangay=Pinagtipunan,
Barangay=Sampalucan, Barangay=Pasong Camachile I, Barangay=96th,
Barangay=Bagumbayan, Barangay=Bacao II, Barangay=Panungyanan,
Barangay=Sta. Clara
Cluster 1: Barangay=Bacao I, Barangay=Tapia
Cluster 4: Barangay=Prinza, Barangay=San Francisco, Barangay=Buenavista III,
Barangay=San Juan II, Barangay=Pasong Camachile I, Barangay=Dulong Bayan,
Barangay=Buenavista II, Barangay=Vibora, Barangay=Sampalucan,
Barangay=Panungyanan, Barangay=Corregidor, Barangay=96th,
Barangay=Pasong Kawayan I, Barangay=Pinagtipunan, Barangay=Sta. Clara,
Barangay=Pasong Kawayan II
Cluster 3: Barangay=Tejero, Barangay=Javalera, Barangay=San Juan I
Cluster 2: Barangay=Manggahan, Barangay=San Francisco,
Barangay=Pasong Kawayan I,Barangay=96th
Cluster 5: Barangay=Alingaro, Barangay=Biclatan, Barangay=Pasong Camachile II,
Barangay=Pasong Camachile I, Barangay=Dulong Bayan, Barangay=Navarro,
Barangay=Corregidor, Barangay=Bacao II, Barangay=Buenavista II,
Barangay=Vibora, Barangay=Buenavista I
Cluster 7: Barangay=Santiago,, Barangay=Arnaldo, Barangay=Panungyanan,
Barangay=Vibora, Barangay=San Gabriel, Barangay=Bacao II,
Barangay=Sampalucan, Barangay=Pasong Camachile I, Barangay=Pinagtipunan,
Barangay=Navarro, Barangay=96th, Barangay=Bagumbayan,
Barangay=Buenavista II, Barangay=Pasong Kawayan II,
Barangay=Sta. Clara, Barangay=Buenavista I
Rule Set for Year 2016
Cluster 6: Barangay=Sta. Clara, Barangay=San Juan II, Barangay=Bacao II,
Barangay=96th, Barangay=Pasong Kawayan I,Barangay=Buenavista I,
Barangay=Bagumbayan, Barangay=Buenavista II, Barangay=Pinagtipunan,
Barangay=Buenavista III, Barangay=Navarro, Barangay=Dulong Bayan,
Barangay=San Juan I, Barangay=Biclatan, Barangay=Corregidor,
Barangay=Arnaldo, Barangay=Pasong Kawayan II, Barangay=Prinza,
Barangay=Pasong Kwayan I, Barangay=Pasong Camachile I,
Barangay=Sampalucan, Barangay=Governor Ferrer, Barangay=Santiago
Cluster 1: Barangay=Alingaro, Barangay=San Gabriel, Barangay=Manggahan
Cluster 4: Barangay=Tejero, Barangay=Pasong Kawayan II, Barangay=San Francisco,
Barangay=Navarro, Barangay=Santiago, Barangay=Governor Ferrer,
Barangay=San Juan II, Barangay=Biclatan, Barangay=Prinza, Barangay=Arnaldo,
Barangay=Buenavista II, Barangay=Pasong Camachille I,
Barangay=Buenavista III, Barangay=San Juan I, Barangay=Corregidor,
Barangay=Pinagtipunan, Barangay=Dulong Bayan
Cluster 2: Barangay=Javalera, Barangay=Bacao I
Cluster 5: Barangay=Pasong Camachile II, Barangay=Pasong Camachile I,
Barangay=Santiago, Barangay=Corregidor, Barangay=96th, Barangay=Prinza,
10. International Journal of Information Technology, Control and Automation (IJITCA) Vol. 7, No.2, April 2017
10
Values Probability Values Probability Values Probability
MonthComttd March 11% March 13% March 16%
MonthComttd February 10% December 10% September 13%
MonthComttd May 10% May 10% July 12%
MonthComttd July 9% January 10% January 8%
MonthComttd April 9% July 8% August 8%
MonthComttd January 9% February 8% October 7%
MonthComttd June 8% October 8% April 7%
MonthComttd November 7% April 8% November 7%
MonthComttd October 7% November 7% May 6%
MonthComttd September 6% June 7% February 5%
MonthComttd December 6% August 6% June 5%
MonthComttd August 6% September 5% December 5%
2016
Population (All)
Variables
2014 2015
Barangay=Sampalucan, Barangay=Governor Ferrer, Barangay=San Francisco,
Barangay=Bagumbayan, Barangay=Biclatan, Barangay=Bacao II,
Barangay=Pasong Kawayan I, Barangay=Sta. Clara, Barangay=Dulong Bayan,
Barangay=Buenavista I, Barangay=Buenavista III, Barangay=Pasong Camachille I,
Barangay=Pasong Kwayan I, Barangay=Navarro, Barangay=Pinagtipunan,
Barangay=San Juan I
Cluster 3: Barangay=Panungyanan, Barangay=Tapia, Barangay=Vibora
Table 8.0 : Cluster Characteristics of Municipality 2 by Seasonal (Month) Year 2014 – 2016
Table 8.0 shows the cluster characteristics of Municipality 2 by Seasonal (Month) 2014 – 2016.
The summary shows that March ranked the highest in the year 2014 - 2016 with 11%, 13%, and
16% probability.
Rule Set for Year 2014
Cluster 5: MonthComttd=August, MonthComttd=October, MonthComttd=January,
MonthComttd=November, MonthComttd=September, MonthComttd=December
Cluster 6: MonthComttd=September, MonthComttd=June, MonthComttd=December,
MonthComttd=November, MonthComttd=August
Cluster 1: MonthComttd=September, MonthComttd=June, MonthComttd=December,
MonthComttd=November, MonthComttd=August
Cluster 2: MonthComttd=March, MonthComttd=December, MonthComttd=October
Cluster 3: MonthComttd=February, MonthComttd=January
Cluster 4: MonthComttd=April, MonthComttd=October, MonthComttd=August,
MonthComttd=June
Rule Set for Year 2015
Cluster 6: MonthComttd=July, MonthComttd=June, MonthComttd=November,
MonthComttd=September, MonthComttd=October
Cluster 1: MonthComttd=January, MonthComttd=December
Cluster 5: MonthComttd=August, MonthComttd=February, MonthComttd=June,
MonthComttd=July, MonthComttd=October, MonthComttd=November,
MonthComttd=September
Cluster 2: MonthComttd=March, MonthComttd=August
Cluster 4: MonthComttd=May, MonthComttd=November, MonthComttd=October
Cluster 3: MonthComttd=April, MonthComttd=February,
MonthComttd=September,MonthComttd=October
Rule Set for Year 2016
Cluster 1: MonthCom=July, MonthCom=September
Cluster 5: MonthCom=January, MonthCom=May, MonthCom=June,
MonthCom=February, MonthCom=October
Cluster 6: MonthCom=November, MonthCom=February, MonthCom=October,
11. International Journal of Information Technology, Control and Automation (IJITCA) Vol. 7, No.2, April 2017
11
Values Probability Values Probability Values Probability
CrimeCategory PHYSICALINJURIES 31% PHYSICALINJURIES 26% MURDER 29%
CrimeCategory THEFT 24% MURDER 21% RAPE 23%
CrimeCategory RAPE 16% THEFT 18% THEFT 18%
CrimeCategory MURDER 14% ROBBERY 11% PHYSICALINJURIES 14%
CrimeCategory ROBBERY 12% RAPE 10% ROBBERY 9%
CrimeCategory CARNAPPING 2% MOTORNAPPING 7% MOTORNAPPING 4%
CrimeCategory HOMICIDE 1% CARNAPPING 5% CARNAPPING 1%
CrimeCategory HOMICIDE 2% HOMICIDE 1%
2016
Population (All)
Variables
2014 2015
Values Probability Values Probability Values Probability
Barangay Halayhay 17% Halayhay 13% Halayhay 25%
Barangay Bagtas 14% Bagtas 13% Calibuyo 11%
Barangay Capipisa 8% Biwas 9% Capipisa 11%
Barangay Bucal 8% Capipisa 8% Bagtas 8%
Barangay Amaya II 7% Bunga 7% Biwas 8%
Barangay Biwas 6% Amaya I 7% Amaya II 7%
Barangay Calibuyo 6% Calibuyo 6% Biga 5%
Barangay Amaya I 5% Bucal 6% Bucal 5%
Barangay Daang Amaya II 4% Amaya VII 4% Daang Amaya I 3%
Barangay Bunga 4% Daang Amaya I 4% Daang Amaya III 3%
Barangay Amaya V 4% Amaya II 4% Daang Amaya II 3%
Barangay Amaya III 3% Amaya III 4% Bunga 3%
Barangay Biga 3% Amaya V 3% Amaya I 2%
Barangay Amaya VI 3% Biga 3% Amaya V 2%
Barangay Daang Amaya III 2% Daang Amaya II 3% Amaya III 1%
Barangay Amaya VII 2% Amaya VI 3% Amaya VI 1%
Barangay Daang Amaya I 2% Amaya IV 2% Amaya VII 1%
Barangay Amaya IV 2% Daang Amaya III 2%
2016
Population (All)
Variables
2014 2015
MonthCom=December, MonthCom=May, MonthCom=June,
MonthCom=January
Cluster 2: MonthCom=March
Cluster 4: MonthCom=April, MonthCom=December, MonthCom=October,
MonthCom=November, MonthCom=May
Cluster 3: MonthCom=August, MonthCom=April,
MonthCom=November, MonthCom=January
Table 9.0 : Cluster Characteristics of Municipality 3 by Crime Category Year 2014 - 2016
Table 9.0 shows the cluster characteristics of Municipality 3 by Crime Category 2014 – 2016.
The summary shows that Physical Injuries ranked the highest in the year 2014 and 2015 with
31%, and 26% probability, and Murder in the year 2016 with 29% probability.
Rule Set for Year 2015
Cluster 3: CrimeCategory=THEFT, CrimeCategory=ROBBERY, CrimeCategory=RAPE,
CrimeCategory=HOMICIDE, CrimeCategory=CARNAPPING
Cluster 2: CrimeCategory=MURDER, CrimeCategory=MOTORNAPPING,
CrimeCategory=CARNAPPING
Cluster 1: CrimeCategory=PHYSICALINJURIES
Rule Set for Year 2016
Cluster 2: CrimeCategory=RAPE, CrimeCategory=PHYSICALINJURIES,
CrimeCategory=MOTORNAPPING
Cluster 1: CrimeCategory=MURDER, CrimeCategory=CARNAPPING,
CrimeCategory=MOTORNAPPING
Cluster 3: CrimeCategory=THEFT, CrimeCategory=ROBBERY,
CrimeCategory=HOMICIDE, CrimeCategory=CARNAPPING
Table 10.0 : Cluster Characteristics of Municipality 3 by Barangay Year 2014 – 2016
12. International Journal of Information Technology, Control and Automation (IJITCA) Vol. 7, No.2, April 2017
12
Table 9.0 shows the cluster characteristics of Municipality 3 by Barangay 2014 – 2016. The
summary shows that Halayhay ranked the highest in the year 2014 - 2016 with 17%, 13%, and
25% probability.
Rule Set for Year 2014
Cluster 5: Barangay=Amaya I, Barangay=Amaya II, Barangay=Amaya VI,
Barangay=Daang Amaya II, Barangay=Calibuyo, Barangay=Amaya V,
Barangay=Daang Amaya III, Barangay=Daang Amaya I, Barangay=Amaya VII,
Barangay=Biga
Cluster 6: Barangay=Amaya III, Barangay=Bunga, Barangay=Biga,
Barangay=Daang Amaya II, Barangay=Daang Amaya III, Barangay=Calibuyo,
Barangay=Amaya IV, Barangay=Daang Amaya I, Barangay=Amaya VI,
Barangay=Amaya II, Barangay=Amaya VII
Cluster 1: Barangay=Halayhay
Cluster 2: Barangay=Bagtas
Cluster 3: Barangay=Biwas, Barangay=Amaya I, Barangay=Amaya IV, Barangay=Calibuyo,
Barangay=Amaya II, Barangay=Bunga, Barangay=Amaya VII, Barangay=Amaya III
Cluster 7: Barangay=Bucal, Barangay=Amaya V, Barangay=Calibuyo,
Barangay=Daang Amaya I, Barangay=Amaya II
Cluster 4: Barangay=Capipisa, Barangay=Amaya III, Barangay=Amaya V, Barangay=Biga,
Barangay=Amaya IV, Barangay=Amaya VII, Barangay=Amaya I
Rule Set for Year 2015
Cluster 7: Barangay=Amaya III, Barangay=Bunga, Barangay=Amaya VII,
Barangay=Amaya VI, Barangay=Amaya V, Barangay=Amaya IV,
Barangay=Biga, Barangay=Calibuyo, Barangay=Daang Amaya II,
Barangay=Amaya II
Cluster 8: Barangay=Bucal, Barangay=Amaya V, Barangay=Amaya II,
Barangay=Daang Amaya II, Barangay=Biga, Barangay=Bunga,
Barangay=Daang Amaya III, Barangay=Amaya VII, Barangay=Amaya VI
Cluster 1: Barangay=Capipisa, Barangay=Amaya I
Cluster 3: Barangay=Halayhay
Cluster 2: Barangay=Bagtas, Barangay=Bunga
Cluster 4: Barangay=Biwas, Barangay=Amaya VI, Barangay=Amaya VII
Cluster 5: Barangay=Daang Amaya I, Barangay=Amaya IV, Barangay=Amaya II,
Barangay=Calibuyo, Barangay=Bucal, Barangay=Daang Amaya II,
Barangay=Biga, Barangay=Daang Amaya III, Barangay=Amaya III,
Barangay=Amaya VII, Barangay=Amaya V
Cluster 6: Barangay=Calibuyo, Barangay=Bunga, Barangay=Daang Amaya I,
Barangay=Amaya VI, Barangay=Amaya II, Barangay=Bucal,
Barangay=Biga, Barangay=Amaya VII, Barangay=Amaya III,
Barangay=Daang Amaya III, Barangay=Daang Amaya II, Barangay=Amaya IV
Rule Set for Year 2016
Cluster 6: Barangay=Amaya II, Barangay=Bagtas, Barangay=Biga,
Barangay=Daang Amaya I, Barangay=Amaya I, Barangay=Bucal,
Barangay=Bunga, Barangay=Amaya V, Barangay=Amaya III,
Barangay=Amaya VI, Barangay=Amaya VII
Cluster 1: Barangay=Halayhay
Cluster 4: Barangay=Biwas,Barangay=Bagtas, Barangay=Daang Amaya III,
Barangay=Daang Amaya II, Barangay=Amaya III, Barangay=Amaya V,
Barangay=Amaya VII, Barangay=Amaya I, Barangay=Daang Amaya I,
13. International Journal of Information Technology, Control and Automation (IJITCA) Vol. 7, No.2, April 2017
13
Values Probability Values Probability Values Probability
MonthComttd December 17% September 12% September 23%
MonthComttd April 13% April 12% August 12%
MonthComttd June 9% November 11% May 11%
MonthComttd January 9% August 10% March 9%
MonthComttd February 9% July 9% July 8%
MonthComttd March 8% February 8% April 8%
MonthComttd August 6% January 8% February 7%
MonthComttd November 6% March 8% October 7%
MonthComttd July 6% December 7% December 5%
MonthComttd October 6% October 6% January 4%
MonthComttd September 6% May 5% June 4%
MonthComttd May 5% June 4% November 3%
2016
Population (All)
Variables
2014 2015
Barangay=Biga, Barangay=Bunga
Cluster 3: Barangay=Calibuyo, Barangay=Biga, Barangay=Daang Amaya II
Cluster 2: Barangay=Capipisa, Barangay=Amaya II, Barangay=Bucal,
Barangay=Amaya VI, Barangay=Amaya VII,
Barangay=Daang Amaya III, Barangay=Amaya I
Cluster 5: Barangay=Bunga, Barangay=Bucal, Barangay=Bagtas, Barangay=Biga,
Barangay=Amaya II, Barangay=Daang Amaya II, Barangay=Daang Amaya I,
Barangay=Amaya I, Barangay=Amaya VI, Barangay=Daang Amaya III,
Barangay=Amaya VII, Barangay=Amaya V, Barangay=Amaya III
Table 11.0 : Cluster Characteristics of Municipality 3 by Seasonal (Month) Year 2014 – 2016
Table 11.0 shows the cluster characteristics of Municipality 3 by Seasonal (Month) 2014 – 2016.
The summary shows that December ranked the highest in the year 2014 with 17% probability,
and September in the year 2015 and 2016 with 12% and 23% probability.
Rule Set for Year 2014
Cluster 6: MonthComttd=August, MonthComttd=July, MonthComttd=September,
MonthComttd=May, MonthComttd=October
Cluster 5: MonthComttd=February, MonthComttd=November, MonthComttd=October,
MonthComttd=July, MonthComttd=August, MonthComttd=May
Cluster 1: MonthComttd=June, MonthComttd=January
Cluster 2: MonthComttd=December
Cluster 3: MonthComttd=April, MonthComttd=September
Cluster 4: MonthComttd=March, MonthComttd=May, MonthComttd=September,
MonthComttd=February, MonthComttd=November
Rule Set for Year 2015
Cluster 4: MonthComtd=June, MonthComtd=July, MonthComtd=February,
MonthComtd=December, MonthComtd=October, MonthComtd=January
Cluster 6: MonthComtd=January, MonthComtd=July, MonthComtd=May,
MonthComtd=February, MonthComtd=October
Cluster 2: MonthComtd=August, MonthComtd=March
Cluster 5: MonthComtd=April
Cluster 3: MonthComtd=September
Cluster 1: MonthComtd=November, MonthComtd=December,
MonthComtd=May, MonthComtd=October
Rule Set for Year 2016
Cluster 5: MonthComtd=December, MonthComtd=June, MonthComtd=April,
MonthComtd=February, MonthComtd=October, MonthComtd=July,
14. International Journal of Information Technology, Control and Automation (IJITCA) Vol. 7, No.2, April 2017
14
Values Probability Values Probability Values Probability
CrimeCategory THEFT 35% THEFT 27% THEFT 40%
CrimeCategory PHYSICALINJURIES 31% ROBBERY 22% ROBBERY 22%
CrimeCategory ROBBERY 15% PHYSICALINJURIES 22% PHYSICALINJURIES 18%
CrimeCategory MURDER 13% MURDER 13% MOTORNAPPING 10%
CrimeCategory RAPE 2% RAPE 8% MURDER 6%
CrimeCategory CARNAPPING 2% MOTORNAPPING 6% RAPE 2%
CrimeCategory HOMICIDE 2% HOMICIDE 2% HOMICIDE 2%
2015 2016
Population (All)
2014
Variables
MonthComtd=November
Cluster 1: MonthComtd=September
Cluster 2: MonthComtd=March, MonthComtd=August
Cluster 4: MonthComtd=January, MonthComtd=February, MonthComtd=October,
MonthComtd=July, MonthComtd=November, MonthComtd=December,
MonthComtd=April, MonthComtd=June
Cluster 3: MonthComtd=May, MonthComtd=April,
MonthComtd=June, MonthComtd=December
Table 12.0 : Cluster Characteristics of Municipality 4 by Crime Category Year 2014 – 2016
Table 12.0 shows the cluster characteristics of Municipality 4 by Crime Category 2014 – 2016.
The summary shows that Theft ranked the highest in the year 2014 - 2016 with 35%, 27%, and
40% probability.
Rule Set for Year 2014
Cluster 2: CrimeCategory=THEFT
Cluster 3: CrimeCategory=ROBBERY, CrimeCategory=MURDER,
CrimeCategory=HOMICIDE,CrimeCategory=CARNAPPING,
CrimeCategory=RAPE
Cluster 1: CrimeCategory=PHYSICALINJURIES
Rule Set for Year 2015
Cluster 3: CrimeCategory=ROBBERY,CrimeCategory=MOTORNAPPING,
CrimeCategory=RAPE,CrimeCategory=HOMICIDE
Cluster 1: CrimeCategory=THEFT
Cluster 2: CrimeCategory=PHYSICALINJURIES
Cluster 5: CrimeCategory=MURDER, CrimeCategory=ROBBERY,
CrimeCategory=MOTORNAPPING
Cluster 4: CrimeCategory=RAPE, CrimeCategory=MURDER,
CrimeCategory=MOTORNAPPING, CrimeCategory=HOMICIDE
Rule Set for Year 2016
Cluster 1: CrimeCategory=THEFT
Cluster 2: CrimeCategory=ROBBERY
Cluster 4: CrimeCategory=HOMICIDE, CrimeCategory=RAPE, CrimeCategory=MURDER,
CrimeCategory=MOTORNAPPING, CrimeCategory=PHYSICALINJURIES
Cluster 3: CrimeCategory=PHYSICALINJURIES, CrimeCategory=MOTORNAPPING,
CrimeCategory=MURDER, CrimeCategory=RAPE
15. International Journal of Information Technology, Control and Automation (IJITCA) Vol. 7, No.2, April 2017
15
Values Probability Values Probability Values Probability
Barangay San Agustin 20% San Agustin 29% San Agustin 36%
Barangay Perez 18% Perez 16% Perez 19%
Barangay De Ocampo 11% Cabuco 10% Luciano 12%
Barangay Inocencio 9% Conchu 9% De Ocampo 6%
Barangay Luciano 9% Osorio 6% Inocencio 5%
Barangay Cabuco 8% Inocencio 6% Cabuco 5%
Barangay Lapidario 6% Luciano 6% Conchu 4%
Barangay Aguado 6% Aguado 5% Lapidario 3%
Barangay Cabezas 3% Lapidario 3% Osorio 3%
Barangay Osorio 3% Cabezas 3% Gregorio 3%
Barangay Lallana 3% De Ocampo 2% Cabezas 2%
Barangay Conchu 2% Gregorio 2% Aguado 1%
Barangay Gregorio 1% Lallana 1% Lallana 1%
2016
Population (All)
Values
2014 2015
Table 13.0 : Cluster Characteristics of Municipality 4 by Barangay Year 2014 – 2016
Table 13.0 shows the cluster characteristics of Municipality 4 by Barangay 2014 – 2016. The
summary shows that San Agustin ranked the highest in the year 2014 - 2016 with 20%, 29%, and
36% probability.
Rule Set for Year 2014
Cluster 3: Barangay=De Ocampo, Barangay=Lapidario, Barangay=Lallana,
Barangay=Conchu, Barangay=Osorio
Cluster 4: Barangay=Cabuco, Barangay=Aguado, Barangay=Cabezas,
Barangay=Osorio, Barangay=De Ocampo, Barangay=Lallana,
Barangay=Conchu
Cluster 2: Barangay=San Agustin
Cluster 5: Barangay=Inocencio, Barangay=Luciano,
Barangay=Aguado,Barangay=Cabezas
Cluster 1: Barangay=Perez
Rule Set for Year 2015
Cluster 2: Barangay=Perez, Barangay=Cabuco, Barangay=Osorio, Barangay=Cabezas,
Barangay=Lapidario, Barangay=Gregorio, Barangay=De Ocampo,
Barangay=Aguado, Barangay=Lallana
Cluster 3: Barangay=Conchu, Barangay=Inocencio, Barangay=Luciano, Barangay=Osorio,
Barangay=Cabuco, Barangay=De Ocampo, Barangay=Lallana, Barangay=Cabezas,
Barangay=Gregorio, Barangay=Lapidario
Cluster 1: Barangay=San Agustin, Barangay=Aguado
Rule Set for Year 2016
Cluster 1: Barangay=San Agustin
Cluster 2: Barangay=Perez
Cluster 3: Barangay=Luciano, Barangay=De Ocampo, Barangay=Conchu,
Barangay=Lapidario, Barangay=Cabezas, Barangay=Aguado,
Barangay=Gregorio
Cluster 4: Barangay=Cabuco, Barangay=Gregorio, Barangay=Cabezas,
Barangay=Osorio, Barangay=Inocencio, Barangay=Aguado,
Barangay=De Ocampo,Barangay=Lallana, Barangay=Lapidario
Cluster 5: Barangay=Inocencio, Barangay=Conchu, Barangay=Lapidario,
Barangay=De Ocampo, Barangay=Osorio, Barangay=Lallana,
Barangay=Aguado, Barangay=Cabuco, Barangay=Gregorio,
Barangay=Cabezas
16. International Journal of Information Technology, Control and Automation (IJITCA) Vol. 7, No.2, April 2017
16
Values Probability Values Probability Values Probability
MonthComttd July 13% July 16% July 18%
MonthComttd December 10% April 13% September 12%
MonthComttd January 10% October 13% May 10%
MonthComttd September 9% August 13% August 10%
MonthComttd May 9% December 8% November 9%
MonthComttd October 8% June 8% January 9%
MonthComttd June 8% November 8% December 8%
MonthComttd April 8% September 5% April 7%
MonthComttd November 7% January 5% March 6%
MonthComttd February 7% February 4% October 6%
MonthComttd August 6% May 4% February 5%
MonthComttd March 6% March 3%
2016
Population (All)
Variables
2014 2015
Table 14.0 : Cluster Characteristics of Municipality 4 by Seasonal (Month) Year 2014 – 2016
Table 14.0 shows the cluster characteristics of Municipality 4 by Seasonal (Month) 2014 – 2016.
The summary shows that July ranked the highest in the year 2014 - 2016 with 13%, 16%, and
18% probability.
Rule Set for Year 2014
Cluster 1: MonthComttd=July, MonthComttd=January
Cluster 5: MonthComttd=February, MonthComttd=March, MonthComttd=June,
MonthComttd=October, MonthComttd=November, MonthComttd=December
Cluster 6: MonthComttd=August, MonthComttd=October, MonthComttd=November,
MonthComttd=February, MonthComttd=December, MonthComttd=March
Cluster 2: MonthComttd=May, MonthComttd=April
Cluster 4: MonthComttd=December, MonthComttd=March, MonthComttd=October,
MonthComttd=November, MonthComttd=June
Cluster 3: MonthComttd=September, MonthComttd=June, MonthComttd=December
Rule Set for Year 2015
Cluster 7: MonthComttd=January, MonthComttd=November, MonthComttd=May,
MonthComttd=March, MonthComttd=September,
Cluster 1: MonthComttd=July
Cluster 5: MonthComttd=December, MonthComttd=September, MonthComttd=June,
MonthComttd=May, MonthComttd=March,
MonthComttd=February, MonthComttd=November
Cluster 6: MonthComttd=June, MonthComttd=December, MonthComttd=February,
MonthComttd=January,MonthComttd=September,
MonthComttd=November, MonthComttd=March
Cluster 4: MonthComttd=April
Cluster 2: MonthComttd=October
Cluster 3: MonthComttd=August, MonthComttd=November
Rule Set for Year 2016
Cluster 3: MonthComttd=November, MonthComttd=December, MonthComttd=April,
MonthComttd=February, MonthComttd=March
Cluster4: MonthComttd=October, MonthComttd=January, MonthComttd=May,
MonthComttd=April, MonthComttd=December
Cluster 1: MonthComttd=September, MonthComttd=August, MonthComttd=January
Cluster 2: MonthComttd=July, MonthComttd=February, MonthComttd=March
17. International Journal of Information Technology, Control and Automation (IJITCA) Vol. 7, No.2, April 2017
17
4.1 APPLICATION SOFTWARE FOR CRIME MANAGEMENT SYSTEM (CRIMS)
The clustering algorithm has been implemented in the application software for Crime
Management System (CriMS) for Philippine National Police – District VI in the province of
Cavite. The following sample screen shots of the system are shown below.
Figure 2.0 Login Page
Figure 2.0 shows the Login Page of the system. The user is asked to input the username and
password in able to access the system.
Figure 3.0 Data Entry
Figure 3.0 shows the data entry of the system. The user inputs the detailed information on the
crime committed such as the date, time, city, barangay, crime category, and etc.
Figure 4.0 Dashboard
Figure 4.0 shows the dashboard of the system which presents the summary of the crime
committed that is total crime per city, total crime per city against person and against property,
total crime per suspect gender, total crime per disposition of the case and total crime over the
year.
18. International Journal of Information Technology, Control and Automation (IJITCA) Vol. 7, No.2, April 2017
Figure 5.0 shows the report analytics. This module presents the totality of the crime committed
per year, city, crime category, city and barangay, and etc. The data can
4.2 . STATISTICAL TREATMENT
The study adapted the ISO 25010:2011 model for system and software engineering quality
requirements and evaluation. The result is interpreted using weighted mean.
Table 15.0 : Software Evaluation (ISO 25010:2011 Quality Model)
International Journal of Information Technology, Control and Automation (IJITCA) Vol. 7, No.2, April 2017
Figure 5.0 Report Analytics
Figure 5.0 shows the report analytics. This module presents the totality of the crime committed
per year, city, crime category, city and barangay, and etc. The data can also be viewed in graphs.
REATMENT
The study adapted the ISO 25010:2011 model for system and software engineering quality
requirements and evaluation. The result is interpreted using weighted mean.
Eq. (5)
Software Evaluation (ISO 25010:2011 Quality Model)
International Journal of Information Technology, Control and Automation (IJITCA) Vol. 7, No.2, April 2017
18
Figure 5.0 shows the report analytics. This module presents the totality of the crime committed
also be viewed in graphs.
The study adapted the ISO 25010:2011 model for system and software engineering quality
Eq. (5)
19. International Journal of Information Technology, Control and Automation (IJITCA) Vol. 7, No.2, April 2017
19
The computed weighted mean for each functional suitability criteria as shown in the table 15.0
have been interpreted using 5 point Likert scale as follows: very functional, highly functional,
functional, poorly functional and not functional at all. As presented in the table, the computed
functional suitability level in all criterions got a mean of 4.4 which can be interpreted that the
functional suitability of the system is highly functional. The overall computed weighted mean for
all criterions is 4.4 which can be interpreted an overall functional suitability level of highly
functional as evaluated by the respondents of the study.
The computed weighted mean for each performance efficiency criteria as shown in the table 15.0
have been interpreted using 5 point Likert scale as follows: very efficient, highly efficient,
efficient, poorly efficient and not efficient at all. As presented in the table, the computed
performance efficiency level in all criterion ranges from 4.4 to 4.5 which can be interpreted that
the performance efficiency of the system is highly efficient. The overall computed weighted mean
for all criterions is 4.4 which can be interpreted an overall performance efficiency level of highly
efficient as evaluated by the respondents of the study.
The computed weighted mean for each compatibility criteria as shown in the table 15.0 have been
interpreted using 5 point Likert scale as follows: very compatible, highly compatible, compatible,
poorly compatible and not compatible at all. As presented in the table, the computed
compatibility level in all criterion ranges from 4.3 to 4.5 which can be interpreted that the
compatibility of the system is highly compatible. The overall computed weighted mean for all
criterions is 4.4 which can be interpreted an overall compatibility level of highly compatible as
evaluated by the respondents of the study.
The computed weighted mean for each usability criteria as shown in the table 15.0 have been
interpreted using 5 point Likert scale as follows: very usable, highly usable, usable, poorly usable
and not usable at all. As presented in the table, the computed usability level in all criterion ranges
from 4.3 to 4.5 which can be interpreted that the usability of the system is highly usable. The
overall computed weighted mean for all criterions is 4.5 which can be interpreted an overall
usability level of very usable as evaluated by the respondents of the study.
The computed weighted mean for each reliability criteria as shown in the table 15.0 have been
interpreted using 5 point Likert scale as follows: very reliable, highly reliable, reliable, poorly
reliable and not reliable at all. As presented in the table, the computed reliability level in all
criterion ranges from 4.3 to 4.5 which can be interpreted that the reliability of the system is highly
reliable. The overall computed weighted mean for all criterions is 4.4 which can be interpreted an
overall reliability level of highly reliable as evaluated by the respondents of the study.
The computed weighted mean for each security criteria as shown in the table 15.0 have been
interpreted using 5 point Likert scale as follows: very secured, highly secured, secured, poorly
secured and not secure at all. As presented in the table, the computed security level in all
criterion ranges from 4.4 to 4.7 which can be interpreted that the security of the system is highly
secured The overall computed weighted mean for all criterions is 4.5 which can be interpreted an
overall security level of very secured as evaluated by the respondents of the study.
The computed weighted mean for each maintainability criteria as shown in the table 15.0 have
been interpreted using 5 point Likert scale as follows: very maintainable, highly maintainable,
maintainable, poorly maintainable and not maintainable at all. As presented in the table, the
computed maintainability level in all criterion ranges from 4.3 to 4.5 which can be interpreted
that the maintainability of the system is highly maintainable. The overall computed weighted
mean for all criterions is 4.4 which can be interpreted an overall maintainability level of highly
maintainable as evaluated by the respondents of the study.
The computed weighted mean for each portability criteria as shown in the table 15.0 have been
interpreted using 5 point Likert scale as follows: very portable, highly portable, portable, poorly
portable and not portable at all. As presented in the table, the computed portability level in all
20. International Journal of Information Technology, Control and Automation (IJITCA) Vol. 7, No.2, April 2017
20
criterion ranges from 4.3 to 4.5 which can be interpreted that the portability of the system is
highly maintainable. The overall computed weighted mean for all criterions is 4.4 which can be
interpreted an overall portability level of highly portable as evaluated by the respondents of the
study.
The Table 15.0 presented the average computed weighted mean for all criterions is 4.4 which can
be interpreted that the system has an overall rating of high quality.
5. CONCLUSION
The study aspire to measure the performance efficiency of the Philippine National Police District
VI in the Province of Cavite using Data Envelopment Analysis and conduct data mining using a
clustering algorithm to identify the crime pattern such as hotspots, hot place, seasonal and
frequency. The results of the performance efficiency were obtained by Municipality 3 as efficient
Decision-Making Unit (DMU). Therefore, inefficient DMUs can benchmark the best practices of
the efficient DMU in its strategic, tactical and operational undertakings. The Clustering
Algorithm was implemented in the web-based software application Crime Management System
(CriMS) developed by the researchers. The researchers recommend that the software application
must be applied in the Philippine National Police District VI in the Province of Cavite. Auxiliary
studies must be conducted to determine the usefulness and effectiveness of the application
software, hence institutionalized its implementation. Supplementary studies should be conducted
by using the results of this study's data set for a detailed numerical benchmarking, empirical
analysis and develop a new algorithm which are functional solution mitigation initiatives to crime
prevention and other methods.
REFERENCES
[1] Mojares, J. G. (2013). Urbanization and its Effect in the Calabarzon Area, Philippines. Journal of
Global Intelligence & Policy,6(10)
[2] Guanghua Wan & Iva Sebastian (2011). Poverty in Asia and the Pacific: An Update. Asian
Development Bank Economics Working Paper No 267
[3] Boussofiane, A., Dyson, R. G., & Thanassoulis, E. (1991). Applied data envelopment analysis.
European Journal of Operational Research, 52(1), 1-15
[4] Norman, M., & Stoker, B. (1991). Data envelopment analysis:the assessment of performance.
John Wiley & Sons, Inc.
[5] Lertworasirikul, S., Fang, S. C., Joines, J. A., & Nuttle, H. L. (2003). Fuzzy data envelopment
analysis (DEA): a possibility approach. Fuzzy sets and systems, 139(2), 379-394
[6] Charnes, A., Cooper, W. W., & Rhodes, E. (1978). Measuring the efficiency of decision making
units. European journal of operational research, 2(6), 429-444
[7] Berry, M. J., & Linoff, G. (1997). Data mining techniques: for marketing, sales, and customer
support. John Wiley & Sons, Inc.
[8] Han, J., Pei, J., & Kamber, M. (2011). Data mining: concepts and techniques. Elsevier.
[9] Agrawal, R., & Srikant, R.(2000, May). Privacy-preserving data mining. In ACM Sigmod Record
(Vol. 29, No. 2, pp. 439-450). ACM
[10] L Rokach, O Maimon (2010). Data mining and knowledge discovery handbook. Springer US.
[11] Sankar Rajagopal (2011). Customer Data Clustering using data mining technique. International
Journal of Database Management Systems (IJDMS ) Vol.3, No.4, November 2011
[12] Chan, S., & Karim, M. A. (2016). Financial market regulation, country governance, and bank
efficiency: evidence from East Asian countries. Contemporary Economics, (1), 39.
doi:10.5709/ce.1897-9254.197
[13] Kinachi, H., Najjari, V., & Alp, İ. (2016). Using Data Envelopment Analysis and Stochastic
Frontier Analysis Methods to Evaluate Efficiency of Hydroelectricity Centers. Gazi University
Journal of Science, 29(1), 167-176
[14] Osman, I. H., Berbary, L. N., Sidani, Y., Al-Ayoubi, B., & Emrouznejad, A. (2011). Data
Envelopment Analysis Model for the Appraisal and Relative Performance Evaluation of Nurses at
an Intensive Care Unit. Journal Of Medical Systems, (5), 1039
21. International Journal of Information Technology, Control and Automation (IJITCA) Vol. 7, No.2, April 2017
21
[15] Ulucan, A. (2011). Measuring the Efficiency of Turkish Universities Using Measure-Specific Data
Envelopment Analysis. Sosyoekonomi, 14(1), 181
[16] Estrada S., Song H., Kim Y., Namn S., Kang S. (2009). A Method of Stepwise Benchmarking for
Inefficient DMUs based on the Proximity-based Target Selection. Expert Systems with
Applications 36 (2009) 11595-11604
[17] Grubesic, T. H., Wei, R., & Murray, A. T. (2014). Spatial Clustering Overview and Comparison:
Accuracy, Sensitivity, and Computational Expense. Annals of the Association of American
Geographers, 104(6), 1134-1156
[18] Jin-ho, J., & Seung-Ryul, J.(2016). Designing a Crime-Prevention System by Converging Big
Data and IoT. Journal of Korean Society for Internet Information, 17(3),115-128.doi:
10.7472/jksii.2016.17.3.115
[19] Ceccato, V., & Uittenbogaard, A. C. (2014). Space–Time Dynamics of Crime in Transport Nodes.
Annals of the Association of American Geographers, 104(1), 131-150
[20] Mohammad A.Tayebi, Martin Ester, Uwe Glasser, Patricia L.Brantingham (2014).
CRIMETRACER: Activity space based crime location prediction", ASONAM, 2014, 2014
IEEE/ACM International Conference on Advances in Social Networks Analysis and Mining
(ASONAM), 2014 IEEE/ACM International Conference on Advances in Social Networks
Analysis and Mining (ASONAM) 2014, pp. 472-480, doi:10.1109/ASONAM.2014.6921628
[21] Sankar Rajagopal (2011).Customer Data Clustering using data mining technique. International
Journal of Database Management Systems (IJDMS ) Vol.3, No.4, November 2011
[22] Phillips, P., & Lee, I. (2011). Crime analysis through spatial areal aggregated density patterns.
Geoinformatica, 15(1), 49-74. doi:10.1007/s10707-010-0116-1
[23] Semwal, V.B., Singha, J., Sharma, P. et al., Multimedia Tools Appl (2016). doi: 10.1007/s11042-
016-4110-y
[24] Devi & Rajagopalan S.P. (2012). Anaylsis of Customer Behavior using Clustering and Assocation
Rule. International Journal of Computer Applications (IJCA) Vol.43, No.23, April 2012
[25] Seddawy, A. E., Khedr, A. & Sultan, T. (2012). Enhanced K-mean Algorithm to Improve Decision
Support System under Uncertain Situation. International Journal of Modern Engineering Research
(IJMER) Vol.2, No.3, Aug 2012
AUTHORS
MS. MENGVI P. GATPANDAN is on her Dissertation of Doctor in Information
Technology at AMA Computer University. She finished her Bachelor of Science in
Computer Science in 1996 at AMA Computer University, Makati City where she
also received the degree of Master of Science in Computer Science in 2003. She is
a full-time faculty of Jose Rizal University for 11 years, and also an active member
of Philippine Society of Information Technology Educators. She is in the academe
in teaching profession for 18 years to present. With a stint in database
management, she is an IBM DB2 and Cognos BI Certified and also holder of NC II
Tesda Certification Program.
DR. SHANETH C. AMBAT received her Doctor of Philosophy in Engineering
major in Information Technology in Hannam University, South Korea in 2009.
She received her Bachelor of Science in Computer Science in 1995 and Master of
Science in Computer Science in 2004 respectively at AMA Computer University,
Makati City. She is currently the Dean of the School of Graduate Studies at
AMA Computer University, Quezon City. Her research interest includes data
envelopment analysis, data mining, SOM, reinforcement learning algorithm, and
fuzzy logic.